Research On A Remote-sensing Geochemistry Nonlinear Inversion Model Based On ETM+ Data In Lalingzaohuo Region

Geochemical exploration delineates the geochemical anomaly area from geochemical data by using mathematical statistics methods.It is one of the preferred techniques in the field of mineral exploration for its rapid,economical and efficient characteristics.The delineation of geochemical anomaly area from geochemical data can provide direction for further prospecting tasks.It can be seen that geochemical data plays a vital role in the process of mineral exploration.However,large scale geochemical data are extremely difficult to obtain where the natural conditions are poor and manpower is tough to reach.With the rapid development of remote sensing technology,obtain of remote sensing data in large areas become more and more easy.Meanwhile,extensive surface geological information can be obtained rapidly with remote sensing data,thus,it have been widely used in geology field.In order to solve the problem of insufficient geochemical data acquired in some areas,a remote sensing geochemical nonlinear inverse model is constructed based on ETM + data in this paper.There is a large amount of redundant data that is not relevant to the model inversion in the traditional band ratio with remote sensing data.Redundant data not only increase the complexity of the inversion model,but also reduce the performance of inversion.The partial least squares regression method and the cross validation strategy are used to extract the most effective components from the original remote sensing data,in order to reduce the redundant information and decrease the computational complexity.Secondly,geochemical sample data expresses the average geochemical information in the corresponding sampling grid.According to this characteristic,a point-to-surface pattern between geochemical sample data and remote sensing image has been constructed for conforming to the real data organization structure.Finally,an extreme learning machine-based remote sensing geochemical nonlinear inversion model has been proposed.In this paper,the Lalingzaohuo region was selected as the research area.Based on the above-mentioned inversion model,three elements,i.e.,iron,molybdenum and copper in the research area were analyzed in this paper.The experimental results are analyzed from experimental verification and practical application to evaluate the effectiveness of the model.In the experimental verification,we can find that the average absolute prediction error of point-to-surface pattern is less than the point-to-point pattern,which can explain the validity of the point-to-surface pattern.We also can find that extreme learning machine is better at predicting compared with partial least squares regression.The compared graph between the results of prediction and real data and the prediction error graph show that the inversion model has higher inversion prediction ability.In practical application,the comparison of delineated anomaly target areas between the inversion data and the original data indicates that the geochemical data acquired by the inversion model has a stronger ability in anomaly recognition.The corresponding relationship between the delineated anomaly target area of the inversion data and the known mineralized spot is better than that of the original geochemical data.Moreover,some unexcavated anomaly areas using original geochemical data can be found in the inversion results.Therefore,it also provides the direction for further prospecting target.